Truth and Predication

Is That All There Is?

Donald Davidson is not widely known to the public, but he was one of the most influential philosophers of his time. He died in 2003. He was working on this book at the time of his death, and it has now been seen through publication by his widow. It is a collection of lectures (the first three being his well-known Dewey lectures, delivered at Columbia University in 1989) and although there is a thematic unity between them, they can also be read profitably as individual essays. The academic essay was Davidson's preferred form; he produced no monograph or book. His many influential papers form his philosophical legacy. They tended to be uncompromising in their difficulty, and unfortunately those in this collection also presuppose a good deal of philosophical background. And yet Davidson's work has a claim on more general attention. Richard Rorty has compared Davidson to Wittgenstein, and predicted that a couple of centuries from now historians of philosophy will be writing about the changes in the human selfimage that he brought about.

An air of profundity enfolded in forbidding technicality: this was Davidson's trademark. Back in the 1970s, his name was everywhere. In many circles he was not only the most influential living philosopher, he was something like a one-man church. Born-again disciples would abandon hitherto cherished problems and projects in order to sign up for baptism. To have attended his lectures, or the lectures or seminars of his few accredited apostles, became a necessary sign of grace. Unbelievers kept a low profile, and were met not so much with arguments as with the pitying contempt that the initiated bestow on the unconverted.

Few had the temerity to ask what the Davidsonian fervor was all about. It was known that the church had a sacrament called Convention T, reverently worn as an amulet, sovereign when shaken at heresy or doubt, and itself derived from the Polish prophet Alfred Tarski. But beyond that not very much was clear. And Davidson's writings afforded no easy points of entry. Still, the Davidsonic boom, as it was irreverently called, continued to sound, and it has continued to grow in volume.

Davidson certainly made serious and imaginative contributions to the philosophy of mind, where a position that he first put forward, known as "anomalous monism," is now a salient landmark. It was provoked by the difficulty of understanding "the efficacy of the mental," or how the mind can affect the brain and body. This in turn Davidson interpreted as the difficulty of holding together three thoughts. The first is that mental events indeed cause physical events: it is because you recall your ineptitude that you blush, or because you want milk that you visit the kitchen. The second is that where there is causation, there is law-like connection: one kind of event causes another when there is a law of nature linking them. The third is that there is nothing law-like about the mental and the physical. There is no law that remembering ineptitude makes people blush, or wanting milk makes them go into a kitchen. Often it depends on what else is in a person's mind.

These three thoughts are obviously hard to hold together, and at first sight they seem inconsistent. Davidson's ingenious solution was to suggest that mental events are nothing but physical events (hence, monism), but that the mental description imports a vocabulary for describing them that is unsuited for occurrence in physical laws (hence, anomalous). There will be a law connecting the neurophysiological event that is the memory of ineptitude with the physiological changes that issue in a blush. But the law will be framed in terms of cellular changes, transmitted impulses and changing chemicals, not in terms of memory or ineptitude. This ingenious suggestion ramified into general accounts of the way in which the mental causes the physical, and hence the question of whether reasons are causes; and this in turn affects how we think of explanation in psychology and history, and eventually how we are to think of ourselves as part of the natural world.

This idea illustrated Davidson's undoubted capacity to take an old problem, turn it around in an original way, and bring out far-ranging and ambitious implications of his new approach. It also, some would say, illustrated something less satisfying. There is room for doubt whether the new solution quite matches up to the original problem. The efficacy of the mental may seem not so much protected as lost in Davidson's account, for the fact that the neurophysiological event is the memory of ineptitude (and what kind of fact is that?) evidently plays no role in the ensuing train of changes. Or so critics argued.

But the lectures in this posthumous collection concern Davidson's other central interest, the philosophy of language, the part built on Convention T as the rock on which the church was founded. What this means can be introduced with a toy example. How does the Roman system of counting work -- the system using symbols like I and II, but also V and X and C and D and M? As beginners, we just learn it. We "cotton on" after a time, and then when a book has "MDCCXLIV" on its title page, we know its date. There are rules, sure enough: it is not just accidental or inexplicable that this sequence of letters refers to 1744. Yet it is not easy to write down the rules explicitly. If we wanted to do so -- say, in order to program a computer to tell us in the ordinary Arabic notation what any sequence of Roman numerals stands for -- we might start by saying what the individual letters stand for (the letter "I" in Roman stands for 1, "V" in Roman for 5, and so on), and then find a way of expressing the result of putting them together, so that "IV" stands for 4, and "VI" for 6, and so on. In Roman it is a little complex, since sometimes putting them together ("concatenation," as it is called in the trade) results in subtraction (as in IV) and sometimes addition (as in VI). But it is not very difficult to tease out the system in this as well, and when we have finished we can point proudly to the result, and say that now we know, explicitly, how Roman works.

The same task can be set for our own Arabic notation, and it is a bit easier. There are the ten digits: 0, 1 ... 9. We can write down what each of these refers to. And there is basically one rule, which is that if you concatenate a sequence of digits with another single digit on the right of it, the result is ten times whatever the original referred to, plus whatever the right-hand digit refers to. Applying and re-applying this rule enables you to say explicitly what any numeral expression of whatever length refers to. (This process is called "recursive." One of the nice things about all this is that it gives you a hefty technical air. If you describe the Arabic notation using the very same notation itself, the result is called "homophonic," which also sounds pretty serious. ) And that's how the notation works. That is why it is not a miracle that the sequence 15234 refers to fifteen thousand two hundred and thirty-four. If we had spoken differently -- if, for example, the sign "4" had been used for the number of hands human beings have, instead of the number of legs dogs have -- then the longer numeral would also have referred to something different.

Numerals refer to numbers. That is what they are for. But sentences do not refer to things. Instead they say things that are true or false. So we cannot immediately generalize from our toy example to the sentences of full-blown language. The great logician Alfred Tarski suggested that we could do something parallel, if we could write down a rule that enabled us, faced with any arbitrary sentence that the language enables us to form, to calculate when it is true. The rules would again work by applying and re-applying to the results of concatenating simple components of a language. (As Quine put it, we are chasing truth up the tree of grammar.) And just as we said we would understand Roman when we could calculate what any arbitrary sequence of numerals referred to, so Tarski proposed that we would understand something important about how a language works if we could calculate for any arbitrary sentence under what conditions it is true. You would satisfy Tarski if you had a program enabling you to compute, for any French sentence whatsoever, a result like this: la neige est blanche is true in French if, and only if, snow is white. Tarski called this satisfying Convention T, and the individual results, such as this one for la neige est blanche, are called T-sentences.

Satisfying Convention T sounds like a reasonable goal, for if some sentence escaped your net, then your rules would not enable you to determine under what circumstances it is true, and to that extent French would be outstripping your understanding of it. More impressively, Tarski called this "providing a definition of truth which satisfies Convention T," for that particular language. His choice of words is arguably a pity, since it carries the sinister implication that the definition of truth for one language is going to be different from that of another, and this may ring philosophical alarm bells. As it stands, the innocent activity of finding a parallel, for a language, of what we so easily found for Roman and Arabic, has no such dark implications. If we said that what we had done was to define "reference to a number" for Roman and for Arabic, we would be misleading ourselves if we went on to conclude that reference to a number was "something different" in the Roman and Arabic worlds. What we have done is to discover how they each do it, but what they each do may be just the same, and similarly the truth that snow is white is the same on either side of the Channel, or the Atlantic.

It proved pretty difficult for logicians to provide rules satisfying Convention T, even for quite simple and artificial languages. For good but complicated reasons, Tarski needed to work in an oblique way, and he himself was pessimistic about extending his methods to the languages that we actually use, which contain nasty complications. All this provided an exciting playground for logicians. And their house jargon -- "concatenation" and "recursive," and also "satisfaction," "sequences," "metalanguage," "finitary methods," "hierarchies," "atomic sentences," and many others -- proved irresistible to philosophers, as dazzling as the latest bling and strutted every bit as proudly. By laying down Convention T, Tarski had done something very interesting. The only problem for philosophers was to decide what it was.

Was it -- holy of philosophical holies -- nothing less than a new and scientific answer to Pilate's question, "what is truth?" Not quite, urged caution. Go back to our toy example. Thinking of ourselves as having defined "reference to a number in Roman" would be a bit off-color. From Pythagoras and Plato onward, philosophers have regarded reference to numbers as one of the great mysteries, right up there with Pilate's question. How do we know anything of timeless, changeless, abstract entities such as numbers? Why is that knowledge useful? We would not deserve much of a hearing if we leaped into the marketplace in Athens waving our rules for Roman and Arabic, and advertising them as solutions to those problems. The reason is obvious enough: our rules simply take reference to numbers for granted, both for individual letters or digits, and for sequences of them. Similarly, Tarski's critics insisted that he did not enable us to answer any interesting questions about truth itself. His methods take truth -- and its close cousin, the satisfaction of predicates (such as "is red") by things (red things) -- for granted. He shows only what counts as the answer to questions about the structure of language -- a valuable topic, but a different one.

Tarski himself, writing in the heyday of positivism (his great paper was published in 1933), hinted that he was not sure that there was a "problem of truth" beyond the kind of problem that can be tackled by his methods. This may seem strange. Why should Pilate's question be supposed to go away? The clever idea is that when Pilate asks "what is truth?" we reply: "you tell me." That is, you tell me which sentence you are interested in (for instance, la neige est blanche) and we will tell you what makes it true or what would make it true if it is false, by repeating the T-sentence, derivable from our understanding of the structure of French: "la neige est blanche is true if, and only if, snow is white." End of story. There is nothing further to be said in general about truth. This is called deflationism. It is exactly as if you went to the marketplace in Athens and really did ask Plato and Pythagoras: "You want to know about reference to numbers? OK, give me a numeral." And they give you 3451. "Right," you say -- now a drum roll, as you perform the computation -- "'3451' in Arabic refers to 3451. There is nothing further to say."

The penultimate sentence gives no news that you could not have worked out for yourself as a competent speaker of your language, so the last sentence seems especially outrageous. But Davidson is attracted by this haughty and dismissive attitude, and he shares the most important part of its ideology. The part he shares is that if you say "what makes a sentence true" by giving its T-sentence, you do not mention facts, states of affairs, human minds, what science is fated to agree upon in the long run, human utility, or anything else that traditional theories of truth have brought into the picture. You do not have to repair to those things, since you simply use one sentence to describe when a given sentence is true. Davidson believes that this slays a host of philosophical dragons.

Consider the way Davidson's theory of truth bypasses reference to facts and states of affairs. It has always been tempting to think of true sentences as "corresponding with the facts," and if we take this seriously it suggests some kind of mapping, whereby elements in the one structure, the sentence, correspond to elements in another structure, such as things and their properties, making up facts. But this soon runs into troubles, many of which are the subject of the insightful chapters in this volume on predication. If a sentence is just a concatenation of words, together picturing a fact, how does anything actually get asserted? There is a gulf between saying that Socrates is wise and merely picturing Socrates in some relation to wisdom. And the whole idea of a fact as some kind of structure is doubtful. As Wittgenstein remarked, if facts were structures, then they might have a location or move about, but they do not. So what a relief it is to be able to remove all the difficulties, by simply using one whole sentence to describe when another is true. Then we do not have to say that sentences "correspond with" or refer to anything at all, but give their meaning in the quite different way that Tarski has shown us.

One might protest that if you say "what a numeral refers to" by giving the equivalent of its T-sentence -- "3145" refers in Arabic to 3145 -- you do not invoke abstract entities, Platonic eternal structures, human minds, formulae, proofs, or anything else that traditional theories of number have brought into the picture. Yet surely this is only a temporary respite. After all, we haven't yet asked what your own understanding of the numeral you use to interpret Roman, or Arabic, requires. Perhaps it brings you into the territory of these old dragons. So declaring them slain might be a result of putting the telescope to a blind eye, rather than anything else. Among beleaguered philosophers, though, any excuse for advertising the twilight of the dragons can be tempting.

In the end, Davidson demurs from deflationism, but for a different reason, which introduces the last element needed to understand his philosophical habitat. When we wrote that la neige est blanche is true in French if and only if snow is white, what we said could be construed in one of two ways. First, it might be an empirical remark about an empirically given system of sound and writing, one with a history and long evolution, used by a particular people in a particular place. If the history and the evolution had been just a little different, it might not even have been true. But second, and quite differently, we could construe it as a stipulation, part of the definition of an abstract structure, to be called French, which might or might not be the language actually used by anyone, either on this planet or anywhere else.

The great Rudolph Carnap was the first to be really clear about this distinction, back in 1942, when he called it the distinction between applied semantics and pure semantics. But almost everyone has been muddled about it ever since. Indeed, this collection begins with a long-winded engagement with the difference. Eventually, and in spite of some false steps, Davidson fights his way to the side of the angels, although he is not nearly as clear as Carnap. He recognizes that if you take the second route, empirical questions do not go away. They simply become framed as the question of whether the structure you now have, together with the rules for generating T-sentences for it, is rightly regarded as the actual language of any given group in a given place and time. That is indeed an empirical question -- and solving it is enough to give an interpretation of the group. It is saying which abstract structure, described à la Tarski, deserves to be called the language they actually use. For comparison, go back to Roman. We could stipulate in dozens of ways how letters are to refer to numbers, and give dozens of different formulae for the result of concatenating them. Only one of those zillions of possibilities gives us the abstract structure actually used by the Romans.

Here is where Davidson found his epiphany. Philosophical labor now splits into two. First, explore the abstract structures, after serving a long novitiate under Tarski, and extend his methods to include apparently intractable structures thrown up by natural languages. That is the technical program, which was pursued with limited success by many philosophers of his day. And second, think what is involved in giving the interpretation of a group by associating them with just one such structure. So Davidson urges that there is something further to say, something that Tarski has not told us. In the case of numbers, you need to say what makes it reasonable to interpret any empirically given group as using the Arabic notation. How do we know, for instance, whether they use the symbol "5" to refer to the number of fingers on a hand or the number of hands on an arm? How do we know that the result of concatenation is always calculated to base ten, and not occasionally to base eight? Similarly in the case of language, what can make it reasonable to select just one abstract structure out of all the zillions of possibilities to describe them?

The answer to this might seem to be best given by looking at developmental psychology, for whatever else we know about language, we know that human beings learn it. The trouble is that we do not know that much about how we do it. So perhaps the method is better studied by imagining a self-conscious adult interpreting the speech of others. Hence, closely following his mentor Quine, Davidson dramatized the process in the heroic figure of the radical interpreter, the lone anthropologist without benefit of dictionaries or help from bilinguals, who sets himself the task of interpreting the speech of another people. In this way the topic shifted from truth to meaning and understanding.

Three things seemed clear to Davidson about the method of radical interpretation. It has to be empirical: the interpreter has nothing to go on but what can be observed of the behavior of his target group. It has to be holistic: he will be solving for many elements simultaneously, and a choice made in one place will affect what is to be said in other places. And it has to proceed under a principle of charity: unless the interpreter supposes the group is generally in touch with the world and what is true about it, he cannot proceed. Empirical sobriety, holism, and charity are what it takes to speed interpretation on its way.

An old-fashioned metaphysician or philosopher of mind would question the value of the whole approach. He might say, as the philosopher David Lewis said, that all this might tell us how we determine the facts, but it does not tell us how the facts determine the facts. Davidson may tell us how we manage to salute others as like-minded with ourselves, but that seems a good deal easier and less exciting than discovering how we or others come to be minded in the first place. Reverting once more to our toy analogy, it is not too difficult to see how to set about interpreting some target group as counting in Roman, but that is not yet to understand how we come to think of eternal abstract objects such as numbers, or even whether that is a sensible description of what we actually do, nor is it telling us much about whether in mathematics truth is dependent on proof, or how much of it there is, or indeed anything else. And declaring such questions to be old hat, finally transcended or overcome by a new emphasis on interpretation, seems at best premature, and at worst a naked takeover bid.

In fact, over the last twenty years, skeptics and apostates from the Davidsonian church have recognized that the interpreter's own understanding is a dragon that remains unslain, grinning and mocking this approach to meaning and interpretation. This is actually in addition to the complaint that the whole edifice is a bit rickety, since if meaning is a satisfactory notion, explained by the method, then this undercuts the need for the logical austerity given by insisting on the ubiquity of T-sentences in the first place. Please do not panic if that does not immediately make sense to you, since very few philosophers noticed it either.

But for Davidson even this was just the overture, and easily forgotten after the real rout of the dragons began. For the truly amazing thing about Davidson's philosophy was the amount he managed to extract from his simple thoughts about interpretation, or rather from these simple thoughts coupled with a resolute policy of declaring anything on which they have no purchase to be thereby obsolete -- nothing but old philosophical phantoms. It turns out that if the method of interpretation is declared to be our only avenue to raising and solving questions about meaning and truth, quite a lot becomes obsolete.

All questions become public questions, so no problem of a connection between meaning and consciousness remains. Davidson argues that only interpreters need the concept of belief (to "take up the slack" between how the interpreter sees the world and any divergence on the part of his subject), but that without such a concept thought is impossible. Skepticism is conquered, since the method of charity ensures that an interpreter will always take us to be saying and thinking things that are mostly true. There is no possibility of declaring anyone to be mostly wrong about most things most of the time, and hence we are ourselves are not so deceived, and Descartes's malin génie is finally vanquished. Tarski had denied that semantics was a system for determining that everyone except you and your friends is talking nonsense, but Davidson overcame any such inhibitions, arguing that there can be no divergent conceptual schemes, since if an interpreter is faced with something that he cannot get hold of by the certified methods, then he has to declare it incapable of truth or falsity, or incapable of meaning at all.

In this way the relativism of different ways of thought, different conceptual systems, is abolished. So Davidson affirms the hegemony of contemporary American more triumphantly than Dick Cheney could dream of, since if a way of thought cannot be interpreted in these homely terms, this shows that it was not thinking after all. This marks the death of any notion of a conceptual scheme, since there is only the one way of thought. And empiricism itself is put in its grave, since it works in terms of conceptual schemes and their content, and now the very idea is abolished.

Nobody could say that Davidson's arguments for these grand conclusions were watertight. At their best they depend upon some kind of principle of verification, declaring possibilities we cannot comprehend not to be possibilities at all. Beyond that, they depend upon apparently arbitrary restrictions on what the interpreter might learn, and upon turning a blind eye to detailed objections, or to the residual problem of how the interpreter understands things in the first place. In some cases, such as that of animal thought, there seemed to be little more to the argument than free association of ideas.

So how did the church of Davidson triumph? Philosophers think of themselves as the guardians of reason, intent beyond other men upon care and accuracy, on following the argument wherever it leads, spotting flaws, rejecting fallacies, insisting on standards. This is how we justify ourselves as educators, and as respectable voices within the academy, and even in public life. But there is a yawning chasm between our self-image and our practice. It is in fact a great mistake to think that philosophers gain their followings or their reputations mainly by means of compelling arguments.

The truth is the reverse: when the historical moment is right, people fall in love with the conclusions, and any blemish in the argument is quickly forgiven. The most outright fallacy becomes beatified as a bold and imaginative train of thought; and obscurity actually befits a deep original exploration of dim and unfamiliar interconnections; and arguments that nobody can follow at all become a brilliant roller-coaster ride toward a shift in the vocabulary, a re-formulation of the problem. Follow the star, and the raw edges will easily be tidied up later. The result was nicely put by Bacon nearly four hundred years ago: "The human understanding is not composed of dry light, but is subject to influence from the will and the emotions, a fact that creates fanciful knowledge; man prefers to believe what he wants to be true."

To benefit from this kind of charity, you have to catch the tide at exactly the right moment. Davidson did so. Ever since the 1950s, or even before, the tide had been running his way. Wittgenstein, various Oxford philosophers, Sellars, and Quine had all inveighed against the privacy of meaning, against the conception of mind that gave rise to Cartesian skepticism, and against the myth of the given at the heart of empiricism. Even before them, philosophers such as Husserl, Heidegger, and Sartre, and also Collingwood in England, thought that the "intentionality" or "world-directedness" of the mind reversed the priority given to our subjective mental space in the Cartesian and early modern traditions. The time had come for a revolt against the idealist or skeptical dissociation between the mind and its world that had obsessed philosophers for hundreds of years.

This behaviorist (or anti-privacy) agenda was part of the spirit of the age, and many philosophers besides Davidson were ready to swim along. Davidson was not cynically exploiting this spirit, but he was caught in the tide along with some of his most able contemporaries. And his expository abilities, with the nicely puzzling mix of promised insight and intimidating technicality, fitted him to be the perfect spokesman for the movement, the heir to Wittgenstein and Quine. By harnessing all these energies, and decorating the result with the incense of serious logic, Davidson could found his church.

It is too early to say whether the church will last. There are many philosophers who find that there is yet life in quite a few of the ancient dragons. But it is quite possible that Rorty is right, that the church will last two hundred years. After all, it promises philosophical salvation, a world free of dragons. Who would want to resist that?

Simon Blackburn is professor of philosophy at the University of Cambridge. He is the author, most recently, of Truth: A Guide (Oxford University Press).

• Read more about this book

Try four weeks of the New Republic Digital absolutely free For nearly 90 years, the New Republic has provided its readers with an intelligent and rigorous examination of American politics, foreign policy, and culture. Today, we're proud to offer a faster, easier, and more economical way to enjoy the magazine — TNR Digital. Subscribe today and we'll give you 4 weeks absolutely free. That's less than 36 cents/week for every word of content available in the print version, a downloadable replica of the print magazine, and an array of special online-only features! Click here to sign up.